Structural Diversity in Alkali Metal and Alkali Metal Magnesiate Chemistry of the Bulky 2,6‐Diisopropyl‐N‐(trimethylsilyl)anilino Ligand

Abstract Bulky amido ligands are precious in s‐block chemistry, since they can implant complementary strong basic and weak nucleophilic properties within compounds. Recent work has shown the pivotal importance of the base structure with enhancement of basicity and extraordinary regioselectivities possible for cyclic alkali metal magnesiates containing mixed n‐butyl/amido ligand sets. This work advances alkali metal and alkali metal magnesiate chemistry of the bulky arylsilyl amido ligand [N(SiMe3)(Dipp)]− (Dipp=2,6‐iPr2‐C6H3). Infinite chain structures of the parent sodium and potassium amides are disclosed, adding to the few known crystallographically characterised unsolvated s‐block metal amides. Solvation by N,N,N′,N′′,N′′‐pentamethyldiethylenetriamine (PMDETA) or N,N,N′,N′‐tetramethylethylenediamine (TMEDA) gives molecular variants of the lithium and sodium amides; whereas for potassium, PMDETA gives a molecular structure, TMEDA affords a novel, hemi‐solvated infinite chain. Crystal structures of the first magnesiate examples of this amide in [MMg{N(SiMe3)(Dipp)}2(μ‐nBu)]∞ (M=Na or K) are also revealed, though these breakdown to their homometallic components in donor solvents as revealed through NMR and DOSY studies.


Result and Discussion
Synthetic studies At otal of ten new crystalline compoundsw ere prepared in this study.T hese comprise the solvent-free parent sodium and potassium amides [Na{N(SiMe 3 )(Dipp)}] 1 (3) Polyamines PMDETAand TMEDAare popular chelating and solubility-enhancing ligands in alkali metal chemistry; [18] they generally help to deaggregate the parent unsolvated compounds as reflected here, though interestingly deaggregation has not occurred in the hemi-TMEDA solvate 8. [19] Monometallic complexes 1-8 were all synthesised by deprotometallation of the startinga mine N(H)(SiMe 3 )(Dipp) by am etal alkyl reagent. In the lithium and sodium cases this was the relevant metal n-butyl reagent, whereas the lower stability of potassium alkyls necessitateds witching to the more stable silylalkylr eagent KCH 2 SiMe 3 ,w hich unlike the n-butyl reagent does not possess any b-hydrogen atoms, and so avoids possible decomposition by such an elimination reaction. Ruhlandt-Senge [16] previously made [K{N(SiMe 3 )(Dipp)}] (probably as aT HF solvate) by deprotonation of the amine with potassium hydride in THF and used it in situ to generate Group 2b is(amides)b ys alt metathesis with the relevantG roup 2m etal iodide. Anwander [13] used as imilar salt metathesis approacht o make as eries of rare earth metal amide complexes,d uring which he isolated [K{N(SiMe 3 )(Dipp)}] in powder form and crystallised the tris(THF) solvate of [Li{N(SiMe 3 )(Dipp)}] and deter-mined its monomeric structure. In our study we observedn o benefit in making the PMDETA solvates 4 and 5 through the salt metathesis of the lithium amide [Li{N(SiMe 3 )(Dipp)}] with sodium tert-butoxide and potassium tert-butoxide, respectively, as crystalline yields of all the products were about5 0% from both deprotometallation and metathesis methods (Scheme 1). It is worth noting that irrespective of the method employed, 1 HNMR monitoring of the filtrates from reactions olutions showed essentially quantitative conversionso ft he amine to amide in all cases. Both the startingm etal reagent and, where relevant,t he donor solvent employed in these reactions were added in slight excesscompared to the parentamine.
Deprotometallation wasu sed also in the syntheseso fm agnesiates 9 and 10 (Scheme 2). n-Butylsodium and trimethylsilylmethylpotassium were utilised as above to generate the respectiveh eavier alkali metal amide from the amine, which in turn was reactedi naco-complexation approachw ith the nbutylmagnesiuma mide to afford the monoalkyl-bisamidomagnesiates 9 and 10.O btained as crystalline solids in yields of 76 %a nd 63 %r espectively, these can be described as lower order ates, in the sense that their alkali metal to magnesium ratio is 1:1; whereas highero rder ates would have 2:1r atios. Note that these structures were formed independent of the startings toichiometry as reaction mixtures that have 3:1 amide/nBu ratios, as in [Na 4 Mg 2 (TMP) 6 (nBu) 2 ], still afforded 9 and 10 as the main products. To the best of our knowledge these are the first alkali metal magnesiates of 2, 6-diisopropyl-N-(trimethylsilyl)aniline to be synthesised, isolatedf rom solution and crystallographically characterised( vide infra). However,s odium andp otassium Group 13 ate complexes are known for aluminium. [20] Crystallographic characterisation Ta bles S7 and S8 in the Supporting Information list the crystal data for all ten new compounds, whileT ableS9c ompares selected metric data across the series.
In contrast the larger sodium centre in 7 cannot be coordinativelys atisfied by the amide/TMEDA combination so akin to the case of the KP MDETAc omplex 5 dimerisation takes place throughN a···CÀSi interactions to close a( NaNSiC) 2 ring ( Figure 8). Na approaches the anionic amido Na tom more closely (2.2847 (14) )t han the neutral TMEDANatoms (mean length, 2.4668 ), giving rise to at rigonal pyramidal primary NaN 3 coordination (sum of bond angles, 339.088), but as in 5 this crowded coordination brings the metal into close proximity with intramolecular C1 (2.9379(16) )a nd intermolecular C14' (2.869(2) ), making it five-coordinate overall.
Magnesiates 9 and 10 conform to the pattern seen with the other polymeric solids 1, 2 and 8 in requiring the addition of [D 8 ]THF for their dissolution in C 6 D 6 .H ence, the solutions will contain THF solvates. The reduced solubility of the donor-free compounds in hydrocarbon solvents such as methylcyclohexane immediatelys ignalledt hat we were dealing with struc-tures distinct from that of the cyclic hexanuclear sodium magnesiate[ Na 4 Mg 2 (TMP) 6 (nBu) 2 ], which shows excellent solubility in hydrocarbon media. The 1 Ha nd 13 CNMR spectra of 9 show two distinct types of amido group. For example, the former shows two p-CH-Art riplet resonances at 6.85 and 6.64, two CH(CH 3 ) 2 septets at 4.27 and two Si(CH 3 ) 3 singlets at 0.30 ppm and 0.25 ppm. This could be due to two distinct amido groups within the same moleculeo ra lternativelyt wo distinct amidocontaining molecules. The chemical shift of the CH 2 -nBu multiplet appearing at À0.26 ppm indicates that this ligand is more likely to be attached to Mg as it would appear more upfield if attached to the more electropositive metal Na {for example,t he same resonance appearsi nC 6 D 6 solutions of [{Mg(TMP)(nBu)} 2 ]a t0 .05 ppm [27] and [Mg(TMP)(nBu)·IPr]( IPr = 1,3-bis-(2,6-diisopropylphenyl)imidazol-2-ylidene) at À0.42 ppm}. [28] Due to the ambiguity of these spectra we turnedt od iffusion-ordered spectroscopy (DOSY) [29] to attempt to find ar esolution. For an accurate molecular weight (MW) determination we utilised the externalc alibration curve (ECC) approachw ith normalised diffusion coefficients that has recently been developedb yt he Stalke group. [30] This novel approach takes into account the shape of the molecule allowing accurate MW predictions with am aximum error of less than 9%.The beautyoft his approachisthat only one internal reference (that also can be the solvent) is necessary,w hereas previous approachesr equired multiple references. In this case tetramethylsilane was employed. DOSY determined MWs of species in solution were estimated using the diffusion coefficient values for 9 in [D 8 ]THF ( Figure S1, Ta bles S2-S3 in the Supporting Information). Twod istinct diffusion coefficients were obtained consistentw ith two distinct species. The best fit for these were found to be the homometallic THF solvates [([D 8 ]THF) x Mg{N(SiMe 3 )(Dipp)}(m-nBu)] 9a (surprisingly the error when x = 1, À6%,w as found to be much less than that for x = 2, À21 %, ar esult at odds with the aforementioned crystal structure, which contained two THF ligands) and [([D 8 ]THF) 2 Na{N(SiMe 3 )(Dipp)}] 9b with the error from their calculated MWs being À6% and À7%,r espectively.S ignificantly the heterometallic ate 9 would have an error of 36 %i fu nsolvated,o r4 3% if solvated by one THF ligand.C learly the magnesiatet hat crystallisesf rom hydrocarbon/arene solution breaksd own to homometallic species in the presenceo ft he strongL ewis base THF (Scheme 3). [31] Organomagnesium compounds are synonymous with redistributionr eactions, most notably the Schlenk equilibrium. Unsurprisingly,t he potassium 10 a and [ ([D 8 ]THF)K{N(SiMe 3 )(Dipp)}] 10 b with small errors of À4% and + 1%,r espectively,v ersus the calculated MWs. Supporting evidence that 9a and 10 a are the same solution species comes from the close similarity of their chemical shifts [for example: 1 H, CH 2 -nBu at À0.26 ppm and À0.22 ppm, respectively;S i(CH 3 ) 3 at 0.30 ppm in both; 13 C, CH 2 -nBu at 9.3 ppm and 9.0 ppm, respectively;( o-C q -Ar) appearing at 145.1 ppm in both].R uhlandt-Sengea lso reported [(THF) 2 Mg{N(SiMe 3 )(Dipp)}(nBu)] and NMR data are consistent with that foundi n9a and 10 a (e.g.,C H 2 -nBu at 9.4 ppm in neat C 6 D 6 ).

Reactivity studies
Our failure to synthesise at emplate ring complex as [Na 4 Mg 2 (TMP) 6 (nBu) 2 ]i ncorporating the arylsilyl amido ligand [N(SiMe 3 )(Dipp)] in place of TMP reduced our expectation of realising enhanced or special reactivities, butf or completeness we carried out some representative metallation-iodination reactions.T able S10 in the SupportingI nformation compiles those carriedo ut between 4,4-dimethyl-2-phenyl-oxazoline (12 a)a nd each of the following complexes:t he solvent-free parents odium andp otassium amides 1 and 2;t he heavier alkali metal magnesiates 9 and 10;a nd forc omparison, the alkylmagnesium amide [Mg{N(SiMe 3 )(Dipp)}(nBu)] (Mg). Table S11 lists the reactions between N,N-diisopropylbenzamide (12 b) and the same set of potential bases. In both cases the attempted metallations were performed in methylcyclohexane and the subsequent iodine quenches were done in THF.F ull details are provided in the Experimental Section.
Owing mainlyt ot heir poor hydrocarbon solubility, unsolvated 1 and 2 proved unreactive with both of the aromatic substrates even under reflux conditions. Though magnesium bases are generally significantly less reactive than alkali metal bases, heteroleptic [Mg{N(SiMe 3 )(Dipp)}(nBu)] proved effective at deprotonating the oxazoline substrate (the iodo product 13 a was quantitativew hen the metallation was performed under reflux conditions), since it containsabutyl anion as well as an amide andi sm ore soluble in hydrocarbon media. The magnesium base also metallated the slightly more challenging benzamide substrate, but much less effectively (yield of iodo product 13 b,4 2%). In both cases the metallation-iodination operation took place regioselectively at the ortho position in keeping with the directed ortho-metallation (DoM) principle. [9] Interestingly,w hen the alkylmagnesium amide is incorporated within the sodium and potassium ates 9 and 10,t he reactivity towards 12 a diminishes especially when the metallation is per-formed at room temperature dropping from 80 %t o1 0% and 11 %, respectively.Rerunning the metallationsu nder reflux conditions greatlyi mproves the yields of the iodo product 13 a, thought hey still fall short of that obtained by [Mg{N(Si-Me 3 )(Dipp)}(nBu)].O nly when the amounto fa te base is doubled do the yields obtained approach1 00 %. Yields of the iodo product 13 b from reaction of 9 or 10 with the benzamide never reach quantitative even at reflux temperature with the best just over 50 %.
Attempts were made to gain insight into the intermediate metallated complexes prior to the iodination step. Am odicum of success was made in the reaction between "sodiumm agnesiate 9"a nd the oxazoline 12 a.T he reaction solution deposited ac rystalline solid in the magnesium complex [Mg{N(Si-Me 3 )(Dipp)}R] (11), in which Ri sortho-deprotonated 4,4-dimethyl-2-phenyl-oxazoline. Unfortunately X-ray crystallographic studies of 11 revealed ah ighly disordered structure that negates its inclusion here, though NMR studies confirmed its formula. Though monometallic 11 could result from ad isproportionation of bimetallic ate 9,i ti sp ossible that 9 never formed in the hydrocarbon medium, but was in fact am ixture of 1 and [Mg{N(SiMe 3 )(Dipp)}(nBu)],w ith 11 forming as ar esult of the latter deprotonating the oxazoline substrate. However, given that the yield of iodo product 12 a using [Mg{N(Si-Me 3 )(Dipp)}(nBu)] on its own was 80 %a t2 5 8C, but is only 10 %u sing 9 as the base under the same conditions, this suggests that the presence of the sodium amide component significantly inhibits reactivity of the alkylmagnesium amide. Since it was cleart hat these magnesiates do not possess robust bimetallic structures like that of the cyclic hexanuclear sodiumm agnesiate [Na 4 Mg 2 (TMP) 6 (nBu) 2 ], it was deemed not worthwhile to explore their reactivity any further.

Conclusion
This study has uncovered ar emarkable s-block complex structural diversity based on the amide [N(SiMe 3 )(Dipp)] À functioning as al igand on its own or in combination with an n-butyl ligand.T he parent sodiuma nd potassium amides adopt infinite chain structures, with linear or zig-zaga rrangements, respectively,a dding to the relatively few knownc rystallographically characterised unsolvated s-block metal amides. Tridentate PMDETAd eaggregatest he sodium polymer to am onomer and also generates am onomeric lithium amide, but it is not sufficient to coordinatively saturate the Kc entre of the amide, which exists as ac entrosymmetric dimer,w ith dimerisation expressed by al ong KÀCH 3 SiMe 2 intermoleculari nteraction that closes an eight-atom (KNSiC) 2 ring. Reducing the chelating capability of the donors olvent via TMEDAh as little effect on the lithium amide, which remains monomeric, but the sodium amide dimerises by means of Na···CÀSi interactions mimicking the case of the PMDETAp otassium amide. Refusing half am olar equivalent of TMEDAd espite a1 :1 K/TMEDA stoichiometricr atio in the solution, the potassium amide crystallises as ah emi-TMEDAs olvate. Its infinite chain structure has distinct K coordination environments, one occupying aN 4 site, the other Scheme3.Breakdownoft he complexes 9 and 10 in [D 8 ]THF. Chem. Eur.J.2016, 22,1 4968-14978 www.chemeurj.org sandwiched between aryl rings of two Dipp ligands, classifies it as an ovel potassium potassiate.
Both crystalline sodium and potassium magnesiates display a2 :1 amido/butyl stoichiometric ratio in infinite helical chain structures conflicting with the 3:1r atio of the template base [Na 4 Mg 2 (TMP) 6 (nBu) 2 ], whichi nspired this study.L ack of hydrocarbon solubility denied any opportunity for these ates to display special reactivities like that of the templateb ase. THF was neededf or solubility but DOSY studies indicatet hat the donor promotes fragmentation of the magnesiates into homometallic moieties. Ther ing architecture is the key feature behindt he special templating metallation ability of [Na 4 Mg 2 (TMP) 6 (nBu) 2 ], but such architectures provedi naccessible with the [N(Si-Me 3 )(Dipp)] ligand.D espite this disappointment the study is important,a si th as clearly established that the presence of aryl groups can be an inhibiting factor to molecular ring formation as their p-faces can engage intermolecularly with alkali metals to help construct polymericc hains. Futurew ork will focus on circumventing this solubility/structural problem by using bulky lipophilic non-aryl ligands.

Experimental Section
General procedures:A ll reactions and manipulations were performed under ap rotective atmosphere of dry pure argon gas using standard Schlenk tube or glovebox techniques. Solvents were dried by heating to reflux over sodium benzophenone ketyl and distilled under nitrogen prior to use. Methylcyclohexane was distilled over sodium metal and stored with molecular sieves (4 ). Deuterated NMR solvents were degasified and stored over molecular sieves (4 )p rior to use. 2, N,N,N',N'',N''pentamethyldiethylenetriamine (PMDETA) and N,N,N',N'tetramethylethylenediamine (TMEDA) were purchased from Aldrich, dried by heating to reflux over calcium hydride and stored with molecular sieves (4 )u nder nitrogen prior to use. nBuLi (1.6 m in n-hexane) and nBu 2 Mg (1 m in n-heptane) solutions were purchased from Aldrich and titrated prior to use. Trimethylsilyl chloride, tetramethylsilane, sodium tert-butoxide, potassium tertbutoxide, 4,4-dimethyl-2-phenyl-oxazoline, N,N-diisopropylbenzamide and 1,10-Phenanthroline were purchased from Aldrich and used as received. nBuNa, [32] K(CH 2 SiMe 3 ) [33] and [N(H)(Si-Me 3 )(Dipp)] [34] (Dipp = 2,6-iPr 2 -C 6 H 3 )w ere prepared according to literature procedures. NMR spectra were recorded on aB ruker DPX 400 NMR spectrometer,o perating at 400.13 MHz for 1 H, 155.5 MHz for 7 Li and 100.6 MHz for 13 C. 1 Ha nd 13 C{ 1 H} spectra were referenced to the appropriate solvent signal, 7 Li NMR spectra were referenced against LiCl in D 2 Oa t0 .00 ppm. Elemental analyses of the compounds 1-10 were carried out using aP erkinElmer 2400 elemental analyser.F ull characterisation details are given in the Supporting Information.

Reactivity studies
Isolation and characterisation of metallated intermediate 11: Complex 9 was chosen as an example. Freshly prepared n-butylsodium (168.2 mg, 2.1 mmol) was suspended in methylcyclohexane (15 mL) and then 2,6-diisopropyl-N-(trimethylsilyl)aniline (998.0 mg, 4mmol) was added. The resulting beige suspension was stirred for 1h and then commercial nBu 2 Mg (2.1 mL, 1 m solution in n-heptane, 2.1 mmol) was introduced by syringe. The reaction mixture was stirred for an additional 1h.A tt his juncture, 4,4-dimethyl-2-phenyl-oxazoline (12 a;3 41.9 mL, 2mmol) was added by syringe. Next the mixture was heated to reflux for 1.5 ht og ive ay ellow solution, which was allowed to cool down to ambient temperature. After ap eriod of one week, colourless crystals of 11 grew from the reaction mixture. Attempts to analyse the crystalline material by X-ray diffraction studies were unsuccessful due to the highly disordered nature within the structure of 11.T hese were filtered, washed with n-hexane (3 4mL) and dried under vacuum (unrefined yield:2 90 mg, 0.60 mmol, 30 %). The absolute yield was higher since the filtrate contained am ixture of the product 11 and starting material. The reaction was also studied using two and three molar equivalents of complex 9 and one equivalent of the substrate (2 mmol) giving the same compound 11.T he NMR spectra of isolated crystalline 11 are in agreement with a1 : Application of metallated compounds in organic synthesis by electrophilic additionreaction General procedure:T he aryl substrates 4,4-dimethyl-2-phenyl-oxazoline (12 a)a nd N,N-diisopropylbenzamide (12 b)w ere treated with the appropriate metal complex in methylcyclohexane. All reactions were stirred at/for different temperatures/times and substrate:base stoichiometries of 1:1a nd 1:2w ere probed. Following metallation, the corresponding 2-monoiodo derivatives 13 a and 13 b were obtained by in situ reaction with an iodine solution in tetrahydrofuran (1 m)a tÀ78 8C. The reaction mixture was allowed to warm up to ambient temperature over ap eriod of 16 h. As aturated aqueous NH 4 Cl solution was added, followed by saturated aqueous Na 2 S 2 O 3 solution. Extraction of the organic crude with ethyl acetate (3 10 mL), then it was washed with brine (10 mL) and dried over anhydrous MgSO 4 .T he solvent was removed under reduced pressure and the crude reaction product was dissolved in CDCl 3 .1 ,10-Phenanthroline was added as internal standard, the yields being calculated by 1 HNMR spectroscopy (see Ta bles S10 and S11i nt he Supporting Information for details). The aromatic substrate 12 a or 12 b (2 mmol) was added to ar eaction mixture of the corresponding monometallic complexes 1, 2, [Mg{N(SiMe 3 )(Dipp)}(nBu)] (2 mmol) in methylcyclohexane (15 mL) or the mixed-metal complexes 9/10 (2 mmol/4 mmol) in methylcy-clohexane (15 mL/30 mL, respectively). All reactions were stirred at/ for 25 8C/24 ha nd 101 8C/1.5 hr espectively.T he iodination reaction was carried out as outlined above.